With multivariate procedures (multivariate analysis (methods), Abk.: MVA) are examined multivariate distributed statistic variables. One not a variable isolates (univariate distributed), but a cooperating of several variables regarded here at the same time, their dependence structure.

Frequently used multivariate methods are

• Regression analysis
• Diskriminanzanalyse
• Main component analysis
• Factor analysis
• Cluster analysis
• Multi-dimensional scaling
• Logistic involution (Logit model)
• Conjoint analysis
• Multiple involution

The classical procedures are throughout linear models, which make special demands against the used data. So the data should not be asymmetrically distributed outliers-free and. When deviating the data from the demanded structure one manages for example, by removing existing outliers or, about Logarithmieren submits the data of a nonlinear transformation. Exist also often alternative methods, which make iterative won solutions possible.

Frequently used criteria for an optimal solution are

• Distances between points in a multidimensional area. Worth mentioning is here above all the Mahalanobis distance, which one could call roughly simplified square of the Euclidean distance.
• Variances, which are minimized and/or maximized. The variance serves in the communication theory as measure for the information content of data.

The manual computation of multivariate ones procedures is mostly very Therefore these methods experienced their upswing only with the development of the EDP. Also frequently only few data concerning probability distributions lying to reason can be given with computed results.

Examples

Examples of use of multivariate procedures:

• In order to provide and find due to comparisons psychological profiles out, who the most probable author/speaker/author are (Kriminologie, linguistics).
• In order to compare the text of an anonymous author with texts of well-known authors and the most probable author to find (a kind of play of the point specified first).
• DATA Mining: Large data sets in data bases are analyzed on unknown structures. One expects here realizations over cooperating different aspects, for example the consumer expenditures of customers as a function of the social status by finding out similarity structures.
• Development of social tuning processes (political sociology) and the influence of individual participants on it.
• Credit standings of debtors (Diskriminanzanalyse).
• With the security analysis: Which enterprise numbers affect mainly the yield capacity of an (Factor analysis)
• With the search for causes for the ice ages (factor analysis).

Literature

• Baking house, Klaus; Erichson, Bernd; Plinke, Wulff: Multivariate analysis methods. Springer, Berlin 2005. ISBN 3-540-27870-2 (standard work)
• Peter Atteslander and others: Methods of the empirical social research. Gruyter publishing house, 2003, ISBN 3110178176 (basic knowledge)
• Anthony P.M. Coxon/P.M. Davies: The User's Guide ton multi-dimensionally Scaling. London, 1984, (Heinemann Educational Books), ISBN 0435822519 and ISBN 0435822527 (so far best English-language representation)

Articles in category "Multivariate procedures"

We found here 10 articles.

Related Websites

We found here 3 related websites.

• More Multivariate Procedures available in NCSS
  Sample display of the loglinear procedure in NCSS.


• Multivariate Procedures available in NCSS
  MULTIVARIATE PROCEDURES - 1. Multivariate analysis refers to a group of statistical techniques that analyze two or more variables at a time, ...


• Multivariate procedures to increase power of microarray-based gene ...
  Multivariate procedures to increase power of microarray-based gene expression.